husk: Husk data.

In CellProfiler/cytominr: Methods for Image-Based Cell Profiling

Husk data.

Description

husk detects unwanted variation in the sample and removes it from the population.

Usage

husk(
  population,
  variables,
  sample,
  remove_outliers = TRUE,
  epsilon = 1e-06,
  remove_signal = TRUE,
  flatten_noise = TRUE
)

Arguments

population	tbl with grouping (metadata) and observation variables.
variables	character vector specifying observation variables.
sample	tbl containing sample that is used by the method to estimate husking parameters. sample has same structure as population. Typically, sample corresponds to controls in the experiment.
remove_outliers	optional boolean specifying whether to remove outliers. Default is TRUE.
epsilon	optional parameter used in husking to offset eigenvalues to avoid division by zero. Default is 1.
remove_signal	optional boolean specifying whether to husk the signal instead of only scaling it down. Default is TRUE.
flatten_noise	optional boolean specifying whether to flatten the noise instead of scaling it up. Default is TRUE. The parameter is ignored if remove_signal is FALSE.

Value

transformed data of the same class as population.

Examples

population <- tibble::tibble(
  Metadata_pert_name = c(NA, NA, NA, NA),
  Metadata_Well = c("A01", "A02", "B01", "B02"),
  Intensity_DNA = c(10, 20, 12, 32),
  Granularity_DNA = c(22, 20, NA, 32),
  Texture_DNA = c(5, 2, 43, 13)
)
variables <- c("Intensity_DNA", "Texture_DNA")
husk(population, variables, population, epsilon = 1, remove_signal = TRUE)
husk(population, variables, population, epsilon = 1e-5, remove_signal = TRUE)
husk(population, variables, population, epsilon = 1, remove_signal = FALSE)
husk(population, variables, population, epsilon = 1e-5, remove_signal = FALSE)

CellProfiler/cytominr documentation built on July 2, 2023, 6:19 p.m.